基于字典及稀疏表示的遥感图像变化检测方法分析-analysis of remote sensing image change detection method based on dictionary and sparse representation.docxVIP

AbstractChange detection of remote sensing images means to detect the change information by analysising the remote sensing images which is acquired at the same area with different time. With the development of remote sensing technology and image information processing analysis technology, the research on change detection of remote sensing images is becomed an important research direction about remote sensing image, and widely used in social and economic fields, such as land resources monitoring, water monitoring, environmental monitoring, forestry monitoring, agricultural investigating, vegetation cover and meteorological monitoring, disaster monitoring and assessment, urban management planning, even in fields such as military reconnaissance and battlefield estimated and so on.This thesis revolves around how to effectively build difference image and detect change information between two temporal remote sensing images, which completes the following three aspects:Proposed a change detection method of remote sensing images based on region segmentation and Kmeans clustering. First, filter the two temporal images to get a difference map, and extract the regions of interest and a definitely unchanged region by maximum entropy threshold segmentation upon the difference map, and then calculate the features of all regions. Finally, divide the regions of intereste and definitely unchanged region into two classifications according to Kmeans clustering algorithm, then get the final change detection. Test on simulated and real remote sensing images proved that this method can improve the accuracy of segmentation, reduce the number of miss pixels in final change detection, and obtain a better effect of the change detection compared with the similar methods.Proposed a change detection method of remote sensing images based on cross sparse representation with pair dictionaries. First, get the sparse representation of the second temporal image block using a local dictionary construct